Originally Posted by
NichG
I run a superhero game this way basically, but there's the added wrinkle that 'actions' are extended courses of action and not just one-off things. So its less like 'what do you do right now' and more like 'what are you going to be doing, until its resolved?'. This is handled in sets of pairwise interactions (with some ability to intervene in someone else's interaction, but not if you yourself are embroiled in one).
Another very similar setup I used for a more D&D-like game was again that you could call out your actions, but 'Initiative' was a skill in the system that specifically was only used for the special action 'I try to do this before they do that', and it had to be something where the thing you're trying to do is aimed at interrupting or invalidating the thing they're trying to do. But you could wait until some of the consequences of not interrupting had been described if you want.
So like, lets say the DM declares 'the monster attacks this character'. Someone could say 'I teleport them out of the way before the attack lands' which would be an opposed Initiative check. But just 'I attack the monster first' doesn't get to be one - stuff like that would effectively be treated as non-blocking (so you and the monster could attack each-other and kill each-other with your damage, rather than it having to be A then B or B then A). However if for example the resolution of the monster's attack was something like 'the character dies', you could do an Initiative check post-hoc to take the blow in place of that character. It wasn't very formalized what things would qualify and what things wouldn't (and honestly the Initiative skill didn't get used all that much), but I sort of like this view of Initiative as being a special action specifically reserved for interrupting things, rather than the default assumption.